Robust estimation of brain stimulation evoked responses using magnetoencephalography

Abstract Magnetoencephalography (MEG) recordings are often contaminated by interference that can exceed the amplitude of physiological brain activity by several orders of magnitude. Furthermore, activity of interference sources may ‘leak’ into the activity of brain signals of interest, resulting in source estimation inaccuracies. This problem is particularly apparent when using MEG to interrogate the effects of brain stimulation on large scale cortical networks. This technical report offers two contributions. Firstly, using phantom MEG recordings we describe an approach for validating the estimation accuracy of brain stimulation evoked responses. Secondly, we propose a novel denoising method for suppressing the leakage of stimulation related signal into recorded brain activity. This approach leverages spatial and temporal domain projectors for signal arising from prespecified anatomical regions of interest. We highlight its advantages compared to the benchmark - spatiotemporal signal space separation (tSSS) - and show that it can more accurately reveal brain stimulation evoked responses.